Highly-parallelized simulation of a pixelated LArTPC on a GPU

Abstract The rapid development of general-purpose computing on graphics processing units (GPGPU) is allowing the implementation highly-parallelized Monte Carlo simulation chains for particle physics experiments. This technique particularly suitable a pixelated charge readout time projection chambers, given large number channels that this technology employs. Here we present first full microphysical simulator liquid argon chamber (LArTPC) equipped with light and readout, developed DUNE Near Detector. software implemented an end-to-end set GPU-optimized algorithms. algorithms have been written in Python translated into CUDA kernels using Numba, just-in-time compiler subset NumPy instructions. GPU achieves speed up four orders magnitude compared equivalent CPU version. current induced 10^3 pixels takes around 1 ms GPU, approximately 10 s CPU. results are against data from pixel-readout LArTPC prototype.